Vendor JSON feeds are cached 12 h; Yandex & Seznam use static official CIDRs.
The real proof is the IP address: a crawler is trusted only when its IP falls inside the official CIDR ranges published by the vendor. The User-Agent can help, but it is easy to spoof; scrapers often copy UA strings while coming from random networks. If the IP isn’t in the vendor range we flag the request, no matter what the UA says.
Yes, both IPv4 and IPv6 networks are fully supported.
Every request from the live demo UI consumes the same number of credits as a direct API call.